Elsevier Announces 2018 Semantic Web Challenge Winners

AMSTERDAM, October 25, 2018 /PRNewswire/ --

IBM Research team wins leading Knowledge Graph competition 

Elsevier, the global information analytics business specializing in science and health, is pleased to announce the winner of the 2018 Semantic Web Challenge (ScWC). The winner was recently announced at the 17th International Semantic Web Conference held in Monterey County, California, USA. The challenge and allocated prize were sponsored by Elsevier.

     (Photo: https://mma.prnewswire.com/media/774921/SWC_Winner_Anna_Lisa_Gentile.jpg )
     (Logo: https://mma.prnewswire.com/media/754760/Elsevier_Logo.jpg )

The Semantic Web Challenge is a highly-prestigious, and the longest-running, competition fostering scientific progress in the field of artificial intelligence on the web. The semantic web and the use of linked data extends the current human-readable web by encoding some of the semantics of resources in a machine-readable form. One of the more prominent current implementations of semantic web research are known as Knowledge Graphs .

The winner of this year's SWC is:

• "Intelligence Augmented with Humans-in-the-loop," by Alfredo Alba, Anna Lisa Gentile, Daniel Gruhl, Petar Ristoski and Steven Welch from IBM's Almaden Research Center in San Jose, CA, USA.

The 2018 Semantic Web Challenge focused on progress in Knowledge Graph population and validation and was organized by SWC Chairs:

• Dan Bennett, Vice President, Enterprise Data Services, Thomson Reuters, USA;
• Axel-C. Ngonga Ngomo, Professor and Head of the Data Science group at Paderborn University, Germany; and
• Heiko Paulheim, Professor and Chair for Data Science at University of Mannheim, Germany.

"The team from IBM Research used an intriguing mix of machine learning and human-based reinforcement," said Prof. Paulheim.

"This year's challenge was particularly difficult for competitors for two reasons," added Mr. Bennett. "First, we were limited in the training data we could provide challengers, and second, many of the supply chain relationships we were scoring against came from a 'long tail' of publicly-traded companies for which there is limited evidence publicly available."

This year's challenge centered on fact extraction from internet sources to create and validate new relationships within a Knowledge Graph. An open data set of organizations, people and financial entities was provided by the organizers, and competing teams were challenged to find and model supply chain relationships that indicate a supplier/customer relationship between these organizations using open web data to build and train their Knowledge Graphs.